- Text
- Empirical,
- Models,
- Sampling,
- Introduction,
- Methodology,
- Predictive,
- Posterior,
- Likelihoods,
- Frequency,
- Forecasts,
- Predicting,
- Bayesian

Predicting Short-Term Interest Rates: Does Bayesian Model ...

Introduction **Short** Rate **Model**s Data Methodology Sampling Scheme Empirical Application 1 Empirical results ConclusionKey findingsPooling forecasts from different short rate models using BMAyields forecast improvements, particularly for BMA forecastsbased on recent predictive likelihoods.Overwhelming evidence to suggest the BMA based on recentpredictive likelihoods give rise to better forecasts than BMAwhich uses in-sample data to determine posterior modelprobability.The improvement of the predictive likelihood depends on thespecification of the diffusion process (whether it allows for noisearrival process or the variance is levels dependent), which inturn relies on the frequency of short rate data examined.

Introduction **Short** Rate **Model**s Data Methodology Sampling Scheme Empirical Application 1 Empirical results ConclusionSingle Factor **Short** Rate **Model**sdr t = µ(r t)dt +σ(r t)dW tdrift term µ(r t) : α 0 +α 1 r t , α 0 +α 1 r t +α 2 rt 2+α3r and otherstdiffusion process σ(r t) : σr γ t (Chan et al,√1992),β 0 +β 1 r t +β 2 r β 3t (Ait Sahalia,1996) or GARCH(1,1) with levelterm (Brenner et al, 1996)Discrete form∆r t = X it A i △t +ε t√f (Zit , B i )△t,E(ε 2 t |Ω t−1 ) = g(W it G i)

- Page 6 and 7: Introduction Short Rate Models Data
- Page 8 and 9: Introduction Short Rate Models Data
- Page 10 and 11: Introduction Short Rate Models Data
- Page 12: Introduction Short Rate Models Data
- Page 15 and 16: Introduction Short Rate Models Data
- Page 17 and 18: Introduction Short Rate Models Data
- Page 19 and 20: Introduction Short Rate Models Data
- Page 21 and 22: Introduction Short Rate Models Data
- Page 23: Introduction Short Rate Models Data